Welcome back. OpenAI facing some criticism over domestic surveillance concerns as the company signs a new agreement with the Pentagon to deploy its advanced AI systems in classified environments. The two sides creating new language around those concerns, reading, quote, 'AI systems shall not be intentionally used for domestic surveillance of US persons and nationals. The department understands this limitation to prohibit deliberate tracking, surveillance or monitoring of US persons or nationals, including through the procurement or use of commercially acquired personal or identifiable information.' Joining me now is the chief financial officer of OpenAI, Sarah Friar. Sarah, it's great to see you again. Thanks so much for being here this morning.
And of course, we've been talking so much about what went on at Anthropic and Anthropic saying, 'Look, we're not going to even work with the Pentagon. We're afraid of this surveillance.' One of the companies, your company's top research executives, resigned over the weekend over this Pentagon deal, citing concerns over surveillance of the American people. What can you tell us about this and the worry over surveillance?
Sure. Thank you, Maria. Look, Sam, I thought answered this really well last week at the Morgan Stanley conference. He started, and it's something I agree with saying, look, we live in a society that is a miracle. And the reason it's a miracle is largely because of the democratic process. And because of that, we think really three things. Number one, national security decisions must remain with democratic governments. AI companies need to operate within democratic processes. And we think that ultimately an elected government needs to retain this authority over just how powerful these technologies have become and are being used to protect the country. But second, it's going to require collaboration between the labs and the government. And because of that, we felt it was important that we set aside, set in place strong guardrails to make responsible collaboration possible. So in particular, as you've noted, clear red lines on no autonomous weapons, no domestic surveillance, and in fact making sure that some of our people can be deployed into the government fully badged, but forward-deployed engineers from OpenAI to make sure it's happening appropriately. But most importantly, we view this as part of the democratic process.
Yeah. And you think you can do that because I mean, look, Sarah, the truth is, the reality is, there's surveillance in the world. I don't know who to worry about more, you know, the US having surveillance or Beijing because China is having massive surveillance programs throughout the United States and that's no secret. We know that. So, you believe you have come up with a good medium in terms of what can be done and what can't be done in terms of government and OpenAI's partnership.
We do and we have found actually working with the government that they've been very constructive on this and I think you go down the right thought path here on thinking about China. It's vital that the US stay on the frontier. Leadership in AI comes down to compute, comes down to talent, comes down to infrastructure and that's exactly what, for example, the big funding round that we just did is all about today. The US has major structural advantages. We have the leading AI labs. We have the top chip companies. We have the strongest research universities. And we have the deepest capital markets. And when that ecosystem works together alongside the government, innovation accelerates. And the priority has to stay on execution. And that means building the compute backbone. That means deploying AI across the economy. It's making sure that we're driving GDP productivity. And then I think it also is about national security too. When one side has stronger weapons than another, bad things can happen. And so we need to make sure that we in the US have the right access.
Yeah. I mean, on Friday, I made the point, you know, it's really scary to think that there could be domestic surveillance, massive domestic surveillance. But you know what's even scarier? China using domestic surveillance throughout America and deepening that. I mean, Beijing says it's accelerating its military technology and the use of AI on the battlefield. Talk to us about how the Pentagon contract will enhance America's military advantage over China and what specifically you want to see done and not done when it comes to AI uses on the battlefield, Sarah.
Yeah, look, I think first of all to go up a level, AI has become a daily utility in our life, right? We're seeing across society, people are putting AI to work. It's happening in education. It's happening in health care. It's causing incredible breakthroughs in science. We're already seeing that. And the scale just tells the story, right? ChatGPT serves 900 million weekly active users. 50 million of those are subscribing because they see such benefit in that intelligence. Last week, we just deployed our most recent model, ChatGPT 5.4. It's our most capable frontier model and it's really important that that model is available across all of these verticals including the government. Our focus continues to be twofold. One, continuing to push the frontier on what AI is capable of and two, turning it into capabilities and products that people everywhere can benefit from. And we want to make sure this breakthrough in this technology becomes very real for people, real economic value being created in the United States and across the globe.
Yeah. No, I love that and I was going there next in terms of this launch, this new version of ChatGPT that you launched last week, the 5.4 model promising sharper reasoning and improved accuracy, faster responses and improved capability to tackle more complex tasks. Congratulations on the launch, Sarah. But I mean, what I'm trying to understand better is, you know, this idea of autonomous weaponry on the battlefield because I mean, to me, as a naive person in terms of all of this technology, I would prefer to see machines on the battlefield than humans. But the issues that Anthropic brought up was that, you know, the machines can just get too aggressive. I guess...
Look, I think ultimately we're talking about technology that has become incredible in terms of its scale and its abilities. But this comes back to how do we want our society to function. We believe strongly in the democratic process and to that end we believe strongly that you need to work alongside democratically elected governments. We know it's going to need collaboration and at OpenAI we wanted to actually take the baton and run towards this to say okay there is a way to do this within a framework that gets everyone comfortable. It's going to need close cooperation and because of that we want to make sure there are the right guardrails in place and in particular be clear on our red lines. No autonomous weapons, no domestic surveillance. And we think we've gotten to a place with the Department of Defense, with the government on this front, which allows us to keep coming back to why are we doing all of this? We see it as a way to break through technology to create real economic value for the United States, for the world at large, and just some of the outcomes we're seeing at companies, breakthroughs in science, breakthroughs in healthcare, and just even how everyday consumers use this technology, right? Today if I put my blood test into ChatGPT it can give me as good a second opinion as any doctor and that's an amazing thing for people who might live in health deserts for example.
Yeah, it sure is and I love the impact on healthcare. Talk to us about some of the deals you've done, Sarah. OpenAI and Amazon forming this new multi-year strategic partnership to bring new and advanced AI capabilities across the world, invested $50 billion in the company whose models will run on the AWS platform. I mean, this AI infrastructure costs are exploding and we know that. So another question is who ultimately pays for all of this? I mean, is it big tech partners, enterprise, is it consumers? Because we talk every day about the cost of this AI infrastructure and it's going up and up.
